Fundamentally Addressing Bromine Storage ... - ACS Publications

Jul 11, 2017 - Fundamentally Addressing Bromine Storage through Reversible. Solid-State ..... In this imaging mode, the signal contrast scales linearl...
0 downloads 0 Views 7MB Size
Article pubs.acs.org/JACS

Fundamentally Addressing Bromine Storage through Reversible Solid-State Confinement in Porous Carbon Electrodes: Design of a High-Performance Dual-Redox Electrochemical Capacitor Seung Joon Yoo,†,‡ Brian Evanko,§ Xingfeng Wang,∥ Monica Romelczyk,† Aidan Taylor,§ Xiulei Ji,∥ Shannon W. Boettcher,*,⊥ and Galen D. Stucky*,†,§ †

Department of Chemistry & Biochemistry, University of California, Santa Barbara, California 93106, United States Materials Research Laboratory, University of California, Santa Barbara, California 93106, United States § Materials Department, University of California, Santa Barbara, California 93106, United States ∥ Department of Chemistry, Oregon State University, Corvallis, Oregon 97331, United States ⊥ Department of Chemistry & Biochemistry, University of Oregon, Eugene, Oregon 97403, United States ‡

S Supporting Information *

ABSTRACT: Research in electric double-layer capacitors (EDLCs) and rechargeable batteries is converging to target systems that have battery-level energy density and capacitor-level cycling stability and power density. This research direction has been facilitated by the use of redox-active electrolytes that add faradaic charge storage to increase energy density of the EDLCs. Aqueous redox-enhanced electrochemical capacitors (redox ECs) have, however, performed poorly due to cross-diffusion of soluble redox couples, reduced cycle life, and low operating voltages. In this manuscript, we propose that these challenges can be simultaneously met by mechanistically designing a liquid-to-solid phase transition of oxidized catholyte (or reduced anolyte) with confinement in the pores of electrodes. Here we demonstrate the realization of this approach with the use of bromide catholyte and tetrabutylammonium cation that induces reversible solid-state complexation of Br2/Br3−. This mechanism solves the inherent cross-diffusion issue of redox ECs and has the added benefit of greatly stabilizing the reactive bromine generated during charging. Based on this new mechanistic insight on the utilization of solid-state bromine storage in redox ECs, we developed a dual-redox EC consisting of a bromide catholyte and an ethyl viologen anolyte with the addition of tetrabutylammonium bromide. In comparison to aqueous and organic electric double-layer capacitors, this system enhances energy by factors of ca. 11 and 3.5, respectively, with a specific energy of ∼64 W·h/kg at 1 A/g, a maximum power density >3 kW/kg, and cycling stability over 7000 cycles.



INTRODUCTION Electric double-layer capacitors (EDLCs), interchangeably referred to as supercapacitors or ultracapacitors, are capable of storing and discharging energy quickly due to a physical ion adsorption/desorption mechanism in the Helmholtz layer.1,2 Compared to EDLCs using nonaqueous electrolytes, EDLCs with nonflammable aqueous electrolytes are in principle safer, and provide higher power density due to their having a lower ionic resistance.1,3−5 However, the energy density of aqueous EDLCs is limited by the narrow electrochemical potential window of water compared to counterparts with organic electrolytes or ionic liquids.6−8 The grand challenge for aqueous EDLCs is to increase specific energy without compromising specific power and cycling stability. A number of hybrid and/or pseudocapacitive systems have been developed that utilize different charge storage mechanisms in addition to, and/or in place of, electric double-layer capacitance to increase energy density.9−15 © 2017 American Chemical Society

One such approach to enhance the energy density is to replace the inert electrolytes of conventional EDLCs with redox-active electrolytes that enable faradaic charge storage.16−23 Compared with the construction of nanostructured solid-state redox-active electrode materials, liquid-state redox-active electrolytes are easier to prepare and scale up, and should be compatible with the carbon electrodes that are currently mass-produced for commercial EDLCs. In order to design redox-active electrolyte systems, redox couples should exhibit fast and reversible electron transfer and not engage in irreversible side reactions and/or degradation over repeated charge/discharge cycles. Additionally, the crossdiffusion of soluble redox couples that causes low Coulombic efficiency and fast self-discharge must be eliminated. The use of ion-selective membrane separators to mitigate self-discharge has Received: May 4, 2017 Published: July 11, 2017 9985

DOI: 10.1021/jacs.7b04603 J. Am. Chem. Soc. 2017, 139, 9985−9993

Article

Journal of the American Chemical Society

Figure 1. Galvanostatic charge/discharge potential profiles of the asymmetric cells (blue curves) can be divided into contributions from the positive electrode (orange curves) and the negative electrode (purple curves), referenced to the centrally placed Ag/AgCl reference electrode. The cells were charged/discharged at a rate of 2 A/g (based on the mass of positive electrode only) to 1 V, with a 4:1 negative to positive electrode mass ratio; the aqueous electrolytes were (a) 1.2 M KBr, (b) 1 M KBr + 0.2 M MEPBr, and (c) 1 M KBr + 0.2 M TBABr.

occurs when the uncomplexed Br−/Br3− redox couple is used in aqueous redox ECs with activated carbon electrodes. Building upon these findings, we design a high-performance and stable aqueous redox EC. High-energy redox ECs must concurrently utilize a catholyte and an anolyte (dual-redox ECs) to maximize faradaic energy storage.19,23,25,26,36 In aqueous systems, these redox couples should have a standard redox potential close to the oxygen evolution potential (for the catholyte) and close to the hydrogen evolution potential (for the anolyte) to widen the operating voltage of the device, as well as good solution compatibility when intermixed.26 To demonstrate the utility of bromide catholyte and our complexation strategy in such a system, we use a highly soluble ethyl viologen (EV) as an anolyte and investigate the compatibility and comparative Br3− complexing capacity. With the use of readily available tetrabutylammonium bromide (TBABr; n-Bu4NBr) as an additive, a dual-redox EC that utilizes bromide and ethyl viologen (EV) produces a specific energy of ∼64 W·h/kg (at 1 A/g and normalized to the dry mass of both positive and negative electrodes; electrochemical characterization of redox ECs including equations used for calculating device performance are detailed in the Supporting Information (SI)), and maintains stability over 7000 cycles.

been reported,24 but such membranes are costly, and primarily for that reason are not practical for commercial applications. Electrostatic attraction has also been proposed as a mechanism to retain redox couples at the surfaces of oppositely charged electrodes, but was shown to have only a minor effect on the suppression of self-discharge rates.25,26 Halogens (I− and Br−) are promising aqueous redox-active species as they are inexpensive, electrochemically reversible redox couples with high solubility.22,25,27−31 In comparison to iodide, bromide has a higher standard reduction potential (0.81 V vs SCE; E°I3−/I− = 0.3 V vs SCE) that further increases energy density. However, this advantage is offset by the corrosive and volatile nature of bromine generated at the positive electrode. Furthermore, soluble Br3− diffuses to the negative electrode, which causes low Coulombic efficiency and fast self-discharge, as well as possible irreversible oxidation or bromination of the anode or anolyte.32,33 In aqueous bromine flow batteries, asymmetric quaternary ammonium salts such as methyl ethyl pyrrolidinium bromide (MEPBr; 1-ethyl-1-methylpyrrolidinium bromide) or methyl ethyl morpholinium bromide (MEMBr; 4ethyl-4-methylmorpholinium bromide) are commonly used to complex Br2/Br3− as an oily liquid phase.34,35 This complexation reduces the reactivity and vapor pressure of bromine while maintaining a mobile, liquid state for flow-system compatibility. However, this approach does not address the cross-diffusion and poor Coulombic efficiency.32 Standard bromine flow batteries avoid these limitations by storing the charged liquid complex in a separate tank away from the cell stack, but this practice is not feasible for non-flow systems, such as in redox-enhanced electrochemical capacitors (redox ECs). Alternatively, ionexchange membrane separators can be used, but these materials are expensive and require addressing significant sealing challenges to be effective in a practical device. In short, a fundamental need is to store Br3− in non-flow energy storage systems in a manner that (1) reduces the unwanted chemical reactivity and vapor pressure of bromine but at the same time (2) suppresses cross-diffusion and self-discharge by (3) a simple and affordable mechanism. In the present work, we introduce for the first time tetrabutylammonium-induced, reversible solid complexation of the Br−/Br3− redox couple in aqueous redox ECs. Systematic studies show that this solid complex is retained in the pores of the high-surface-area carbon electrodes upon charging, dramatically improving cycling stability and slowing self-discharge relative to traditional oily-liquid-phase bromine complex. Furthermore, we reveal the underlying cause of the irreversible capacity loss that



RESULTS AND DISCUSSION Properties of Br3− Complexing Agents. In order to support the efficient use of the Br−/Br3− redox couple for aqueous redox ECs, we selected and compared two differently structured quaternary ammonium salts as complexing agents for Br3− (complexing agent refers to Br3− complexing agent throughout). Asymmetric cyclic MEPBr, a commonly used complexing agent for aqueous bromine flow batteries, forms a separate oily phase with Br3− while the symmetric and more hydrophobic TBABr is known to form a solid complex.34,37 We studied how each complexing agent affects the overall performance of the redox ECs relative to a control system without either additive. Three asymmetric EC cells were constructed, all utilizing the Br−/Br3− redox couple at the positive electrode, with (1) no complexing agent (1.2 M KBr electrolyte), (2) MEPBr (1 M KBr + 0.2 M MEPBr), and (3) TBABr (1 M KBr + 0.2 M TBABr). The negative counter electrode was purely capacitive and oversized to increase capacity so that the positive electrode could reach the Br− oxidation potential with a total applied cell voltage of 1 V. The influence of each complexing agent on the electrochemical behavior of the bromide catholyte in redox ECs was 9986

DOI: 10.1021/jacs.7b04603 J. Am. Chem. Soc. 2017, 139, 9985−9993

Article

Journal of the American Chemical Society determined by galvanostatic charge/discharge (GCD) cycling tests of each asymmetric cell in a three-electrode configuration (Figure 1). All cells show electrochemical behavior that transitions from capacitive to faradaic energy storage at the positive electrode (orange curve), indicating oxidation of Br− to Br3−, and only a capacitive (linear potential vs time/charge) response at the negative electrode (purple curve) during charging. Notably, there are differences when TBABr is present as a complexing agent as compared to the KBr and KBr/MEPBr cells. For example, a greater potential of zero charge (PZC) was observed for the KBr/TBABr cell (0.26 V) compared to the KBr (0.09 V) and KBr/MEPBr (0.13 V) cells. The more positive PZC with KBr/TBABr electrolyte allows bromide oxidation to occur earlier in the charging stage and increases the ΔV for the capacitive negative electrode, which translates to greater faradaic charge storage at the positive electrode before the arbitrary cell cutoff potential of 1 V is reached (Figure 1; orange curve). Considering the amphiphilic nature of the tetrabutylammonium cation (TBA+),38,39 we reasoned that TBABr acts as a surfactant and TBA+ adsorbs at the porous carbon electrodes. To verify this hypothesis, contact-angle measurements were conducted. The contact angle for the carbon electrode with a droplet of aqueous TBABr solution is ∼24° smaller compared to that with a droplet of aqueous KBr solution (138°; Figure S1). This surfactant behavior may increase wettability of hydrophobic electrodes, and thus facilitate infiltration of hydrophilic bromide catholyte into the porous carbon electrode.39 We also observed that the onset point of the faradaic plateau at the positive electrode is a function of the complexing agent used. The horizontal gray dashed line drawn at 0.7 V in Figure 1 makes the largest negative shift in the Br−/Br3− redox potential for the KBr/TBABr cell. Additional cyclic voltammetry shows a redox potential that is consistent with the above GCD potential profile result, as well as a reversible redox reaction in the presence of TBABr (Figure S2). The observed electrochemical behavior can be explained by the reversible formation of [TBA+·Br3−] solid complex that decreases the Br3− concentration at the interface between the electrolyte and the high-surface-area electrode. Both chemical (Figure 2a) and electrochemical (Figure 2b) methods confirm the generation of a [TBA+·Br3−] solid complex. The image (Figure 2a; right vial) visually shows that after complexation induced by TBA+, Br3− is almost entirely removed from the solution. Notably, the solutions with MEPBr (middle vial) and KBr (left vial) retain a yellow tint, demonstrating that a significant concentration of Br3− remains dissolved due to insufficient complexation by MEP+ and K+, respectively. Figure 2b shows electrochemical generation of a solid complex on the surface of a carbon cloth by passing anodic current in a KBr/ TBABr electrolyte. The Raman spectrum taken of this solid exhibits an intense band at 169 cm−1, which corresponds to symmetric stretching vibration of Br3−,40,41 and its spectrum is identical to the Raman spectrum of commercial [TBA+·Br3−] solid complex (Figure 2c). Further characterization of the electrochemically generated [TBA+·Br3−] solid complex in an aqueous system is well-documented elsewhere.37 Strong adsorption of the charged products within the electrodes has been found to be an effective mechanism to prevent cross-diffusion in a full redox EC cell.26 We hypothesize that our complexation strategy may suppress cross-diffusion even more effectively if the [TBA+·Br3−] solid complex is retained in the pores of the high-surface-area carbon electrodes, instead of being

Figure 2. (a) Complexation of chemically generated Br3− by the addition of potassium bromide (left), MEPBr (middle), and TBABr (right). (b) [TBA+·Br3−] solid complex electrochemically generated by controlled potential electrolysis. (c) Raman spectra of the electrochemically generated [TBA+·Br3−] solid complex (blue curve) and a commercial [TBA+·Br3−] solid complex (orange curve).

formed and precipitating to the bulk solution upon charging. In order to investigate this hypothesis, controlled potential electrolysis with 1 M KBr/0.2 M TBABr solution was performed with an activated porous carbon pellet as the working electrode (denoted as electrolyzed C pellet) to provide an extreme example of the cathode in the charged state. As a control, an identical activated carbon pellet (denoted as the control C pellet) was separately prepared and infiltrated with the same electrolyte solution through repeated vacuum/N2 steps. No electrolysis was performed with this control carbon and hence no [TBA+·Br3−] solid complex should be present in the sample. Both carbon surfaces were analyzed by scanning transmission electron microscopy (STEM) with the signal being acquired using a high-angle annular dark-field detector (HAADF-STEM). In this imaging mode, the signal contrast scales linearly with sample thickness and as the square of average atomic number (Z) of atoms in the sample.42,43 Therefore, for our sample analysis, the higher average atomic number of the bromine-containing [TBA+·Br3−] solid complex compared to that for porous carbon means that such complexes confined in the pores appear as bright regions in the image (Z-contrast imaging). Figure 3a shows a STEM image of the electrolyzed C pellet. The material displays clear contrast with nm-scale bright clusters that were hypothesized to contain the element Br. Subsequent elemental analysis by energy-dispersive X-ray spectroscopy (EDX) confirms the presence of Br in the carbon sample (Table 1). In contrast, the STEM image from the control C pellet displayed no high-contrast spots (Figure 3b), showing that infiltrated Br− 9987

DOI: 10.1021/jacs.7b04603 J. Am. Chem. Soc. 2017, 139, 9985−9993

Article

Journal of the American Chemical Society

these data show that the [TBA+·Br3−] solid complex generated in situ during electrolysis is contained in the pores of the highsurface-area carbon. This complexation approach combined with porous carbon electrodes being used for EDLCs may lead to enhanced stability and slower self-discharge of the bromidebased redox ECs, without using ion-exchange membranes. We note that if a nonporous electrode is used, precipitation of the [TBA+·Br3−] solid complex away from the electrode occurs, which would cause irreversible capacity loss (Figure S3). In order to test stability and self-discharge rates, asymmetric cells were constructed in the two-electrode cell configuration with a paper separator (Whatman no. 1). In cycling stability tests, the KBr cell lost ∼20% of its specific energy over 500 cycles. In contrast, the KBr/TBABr and KBr/MEPBr cells show no fading over the course of 500 cycles, and produce ∼80% and ∼40% higher energy density compared to the KBr cell, respectively (Figure 4a). Large differences between the cells are further evident from self-discharge tests. The self-discharge rate determined by the open-circuit energy efficiency, ηR, of the KBr/TBABr cell was substantially lower than that for KBr and KBr/MEPBr cells, retaining 50% of its energy after 10 h at open circuit, ηR(10 h) = 50%, compared to ηR(10 h) of only 1% and 14% for the KBr and KBr/MEPBr cells, respectively (Figure 4b). Importantly, when the KBr cell was charged to 1.2 V to attain a higher capacity, which is closer to that of the KBr/TBABr cell, an even faster self-discharge rate was observed (Figure 4b, solid blue vs dashed blue). This result implies that the self-discharge problem with uncomplexed Br−/Br3− redox couple is worsened at higher states of charge, limiting its application in highperformance redox EC devices. The problems with cycling stability and self-discharge are much aggravated at 40 °C for the KBr and KBr/MEPBr cells relative to the KBr/TBABr cell, suggesting that the increased diffusivity of soluble Br3− at high temperatures accelerates the redox shuttle self-discharge mechanism (Figure 4c,d; results for self-discharge tests from 0 to 40 °C are summarized in Table S2). These studies demonstrate that charge retention of bromide catholyte, and the stability and self-discharge performances, depend on whether Br3− is uncomplexed, complexed with MEP+ (liquid state), or complexed with TBA+ (solid state). Our results suggest that [TBA+·Br3−] solid complex generated within the electrode pores suppresses cross-diffusion of oxidized bromide and lowers the self-discharge rate. Meanwhile MEPBr, a common complexing agent for bromine flow batteries, shows little improvement over a cell without a complexing agent. The molecular structure of MEP+, which lacks hydrophobic substituents, appears to provide insufficient complexation/ precipitation with Br3−,37 leaving high concentrations of free Br3− in the electrolyte solution (Figure 2a). The question as to whether the decrease in specific energy observed for KBr and KBr/MEPBr cells is temporary due to reversible shuttling/diffusion effects or irreversible fading from chemical reactions was addressed by disassembling tested cells and assessing changes to the structure/morphology of the filter paper separator and carbon-electrode cell components. The separator from the KBr/TBABr cell appeared identical to the pristine paper separator before cycling, while the separators from the other cells are covered with a black solid (pictures in Figure S4). The Raman spectra collected of the separator from the KBr/ TBABr cell shows no clear peaks and is identical to the Raman spectrum obtained from a pristine separator (Figure 5a). However, Raman spectra of the separators from the KBr and KBr/MEPBr cells show peaks at approximately 1340 and 1610

Figure 3. Scanning transmission electron microscopy images of (a) the electrolyzed C pellet and (b) the control C pellet, collected by a HAADF detector. Sample preparation steps including controlled potential electrolysis are detailed in the SI.

Table 1. Elemental Analysis by EDX of the Electrolyzed and Control C Pellets: Relative Atomic Ratio of Carbon to Bromine Present in the Sample (Based on Atom %) sample

C

Br

electrolyzed C pellet control C pellet

28 100

72 0

from KBr/TBABr solution is effectively removed during washing steps. Consistent with the STEM image, elemental analysis by EDX of the control carbon did not detect a Br signal (Table 1). This control experiment strongly supports the hypothesis that the bright regions observed by STEM of the electrolyzed C pellet (Figure 3a) originate from the high-Z insoluble bromidecontaining [TBA+·Br3−] complex that persists after washing, and not by Br− from the solution which is easily removed. In sum, 9988

DOI: 10.1021/jacs.7b04603 J. Am. Chem. Soc. 2017, 139, 9985−9993

Article

Journal of the American Chemical Society

Figure 4. Two-electrode asymmetric cells were constructed with aqueous electrolytes of 1.2 M KBr (blue curve), 1 M KBr/0.2 M MEPBr (black curve), and 1 M KBr/0.2 M TBABr (orange curve). The cells were charged/discharged at a rate of 2 A/g (based on the mass of positive electrode only) to 1 V, with a 4:1 negative-to-positive electrode mass ratio. (a) Cycling stability of each cell tested at 25 °C; the inset shows two-electrode potential profiles vs capacity. (b) Open-circuit (OC) energy efficiency (ηR) for each cell at 25 °C. Each data point was collected by charging the cell, allowing it to sit at open circuit for 2, 6, or 10 h, then discharging the cell completely. For the KBr electrolyte, self-discharge was tested for both a 1 V charge (solid blue) and a 1.2 V charge (dashed blue). (c) Cycling stability of each cell tested at 40 °C. (d) ηR at a given time for each cell at 40 °C.

Figure 5. (a) Raman spectra of the separators after cycling. Raman peaks for the non-cycled carbon electrode (green curve) are added to the plot for comparison. (b−e) SEM images of positive electrodes: (b) Electrode before cycling. (c) Electrode from a KBr cell after cycling. (d) Electrode from a KBr/TBABr cell after cycling. (e) Electrode from a KBr/MEPBr cell after cycling.

cm−1 which can be assigned to the D band and G band, respectively, of activated carbon.44 The same peaks are observed for the non-cycled carbon electrode, indicating that the black solid observed in the separator was shed from the electrode. A control experiment where a pristine separator is soaked in Br2/ Br3− solution (that mimics the charged state of the cell)

demonstrated no color change of the separator in the absence of the carbon electrode. Electron microscopy shows that the electrode exhibited little morphology change when cycled in a KBr/TBABr cell (Figure 5d). In contrast, we observe cracks and partial pulverization of carbon particulates on the electrode surface after cycling in KBr (Figure 5c) and KBr/MEPBr electrolytes (Figure 5e), providing 9989

DOI: 10.1021/jacs.7b04603 J. Am. Chem. Soc. 2017, 139, 9985−9993

Article

Journal of the American Chemical Society evidence of carbon electrode degradation and its subsequent deposition onto the separator (TEM images of the positive electrodes after cycling in Figure S5). Elemental analysis by EDX of the positive electrodes from the KBr and KBr/MEPBr cells after cycling further confirm decreased carbon content (relative to F in the inert PTFE binder) compared to the electrode before cycling (Table S3). Based on the experimental results, we propose that electrochemical intercalation induced by reactive bromine/polybromide causes permanent electrode degradation with time and thus irreversible fading of the KBr and KBr/MEPBr cells.45,46 When TBA+ is present, the concentration of noncomplexed Br2/Br3− is significantly reduced, thereby slowing irreversible side reactions and largely preventing electrode degradation upon repeated charge/discharge cycles. This TBA+-induced solid complexation suppresses the permanent capacity loss mechanism. Overall, these results show an efficient way to utilize the bromide catholyte with readily available TBABr, the simple addition of which simultaneously addresses the capacity loss as well as selfdischarge problems for static bromide-enhanced redox ECs. Designing a High-Performance Viologen/Br DualRedox EC with the Addition of Tetrabutylammonium Complexing Agent. For maximum faradaic energy storage, redox ECs should utilize redox-active electrolytes at both the positive electrode (catholyte) and negative electrode (anolyte) concurrently (dual-redox ECs). We chose viologen (V) as a model anolyte to match the redox activity of the bromide catholyte and test the utility of tetrabutylammonium complexing agent in a full cell. While many families of redox-active anolytes should be compatible with the TBABr/Br− catholyte, viologens have the advantage of an electrochemically reversible V2+/V•+ redox couple.36,47−49 In addition, V/Br cells typically have potential plateaus around 1.2−1.3 V, which coincides ideally with the thermodynamic stability window of water. For the V/Br systems, (1) cycling stability largely depends on effective bromine storage at the positive electrode,36 (2) only viologens with hydrophobic substituents, e.g., pentyl (PV)36 or heptyl viologen (HV),26 can function as an efficient Br3− complexing agent, limiting the capacity for faradaic energy storage due to their low solubility (e.g., < 0.2 M for HV), and thus (3) it is desirable to enhance energy density by using highly soluble viologens to increase the available concentration of anolyte while finding other means to complex Br3− and maintain cycling stability. We envisioned that addition of TBABr to V/Br electrolytes should (1) suppress the fading mechanism at the positive electrode and (2) enable the use of more-soluble, shortalkyl-chain viologens at the negative electrode, in order to (3) maintain cycling stability while producing high energy density in a single device. To verify our hypothesis, readily available ethyl viologen (EV; 1,1′-diethyl-4,4′-bipyridinium dibromide) was selected as an anolyte due to its high solubility (>2 M) and the stability of EV2+/EV•+ redox couple with a highly cathodic redox potential (E1/2 = −0.64 V vs Ag/AgCl) in aqueous electrolytes.50 Additionally, EVBr2 shows good solution compatibility with a bromide catholyte (e.g., NaBr) in an analytical voltammetry cell (Figure S6). First, a control cell with a readily soluble 1.2 M EVBr2/3 M NaBr electrolyte, without additional complexing agent, was assembled and tested. Three-electrode GCD profiles show dual faradaic responses at both positive and negative electrodes, corresponding to Br− oxidation to Br3− and EV2+ reduction to EV•+, respectively, upon charging (Figure 6a). This cell produced a specific energy of ∼68 W·h/kg at 1 A/g, but the cell capacity

Figure 6. 1.2 M EVBr2/3 M NaBr cells. (a) Three-electrode GCD potential profiles for the positive electrode (orange curve), negative electrode (purple curve), and total cell (blue curve) cycled to 1.35 V at 1 A/g. During charging, the electrochemical behavior transitions from capacitive to faradaic at both electrodes. (b) Specific energy, cycling stability, and Coulombic efficiency of an EV/Br two-electrode cell cycled to 1.35 V at 1 A/g.

faded and only retained 82% of the initial energy over 1000 cycles (Figure 6b). This poor cycle life confirms that short-alkyl-chain viologens do not effectively retain reactive Br2/Br3−, which decreases cell stability. To improve the cycle life of EV/Br redox ECs, electrolyte design must utilize a Br3− complexing agent that has stronger Br3− complexing capacity than EV. The degree of (solid) complexation of TBABr with Br3−, relative to EV, was tested by determining how much EV precipitates from an aqueous Br3− solution with and without this complexing agent additive. The amount of precipitated EV was indirectly quantified by measuring the concentration of the remaining viologen in solution with ultraviolet−visible spectroscopy (UV−vis). Figure 7 shows that the addition of Br3− decreased the concentration of EV dissolved in the solution by ca. 83% due to precipitation of the nominally [EV2+·2Br3−] solid complex (EV2+; blue open-square curve vs EV2+/Br3−; blue open-circle curve), which still leaves substantial free, noncomplexed Br3− present in the solution, causing the poor cycling life of the EV/Br cell. Importantly, when TBABr was added together with Br3− and EV, EV remained completely dissolved (EV2+; blue open-square curve vs EV2+/Br3−/TBA+; orange solid curve). These UV−vis results confirm that TBABr has a greater Br3− complexing capacity than EV2+ and that [TBA+·Br3−] solid complex is preferentially formed. Notably, MEPBr, a common Br2/Br3− complexing agent for aqueous flow battery systems, does not 9990

DOI: 10.1021/jacs.7b04603 J. Am. Chem. Soc. 2017, 139, 9985−9993

Article

Journal of the American Chemical Society

strength, i.e., MEP+ < EV2+ < TBA+, is consistent with previous findings that indicate hydrophobicity enhances solid complexation,36,37 and suggests that TBABr should be a suitable complexing agent for EV/Br dual-redox ECs. A two-electrode cell with a 1.2 M EVBr2/2.88 M NaBr/0.12 M TBABr electrolyte (i.e., 10 mol % addition of TBABr based on the molar concentration of EVBr2), paper separator, and symmetric activated carbon electrodes (areal mass loading of 12.7 mg/cm2) was assembled and tested by GCD. The concentrations of EVBr2 and NaBr were set at 1.2 and 3 M, respectively, for the comparison with the standard EV/Br cell, and 10 mol % addition of TBABr ensures that this electrolyte concentration is readily soluble, thus providing simple solution preparation and lower viscosity. The cell (denoted as EV/TBA/ Br) produced ∼64 W·h/kg at 1 A/g with ∼97% Coulombic efficiency and a high energy efficiency of ∼84% when charged/ discharged between 0 and 1.35 V. Most importantly, the specific energy of the cell degraded by only ∼3% over 1000 cycles compared to 18% capacity loss of the cell without TBABr cycled under the same conditions (Figure 8a). For the long-term cycling test, the EV/TBA/Br cell maintained stability with 90% energy retention over 5000 cycles (charge/discharge at 2 A/g) and exhibited ∼4% degradation over an additional 7000 cycles when operated from 0 to 1.3 V at the same current density (Figure 8b). This stability increase shows that the superior performance of short-alkyl-chain viologens can be maintained without sacrificing lifetime through the addition of TBABr. Furthermore, the EV/

Figure 7. UV−vis absorption spectra quantifying the concentration of EV2+ remaining dissolved in the supernatant without Br3− (EV2+; blue open-square curve), with the addition of Br3− (EV2+/Br3−; blue opencircle curve), with Br3− and TBABr (EV2+/Br3−/TBA+; orange solid curve), and with Br3− and MEPBr (EV2+/Br3−/MEP+; black solid curve). The spectra are normalized to the absorption maxima of the EV2+ (λmax = 260 nm). Experimental details for UV−vis measurements are in the SI.

change the degree of solid complexation/precipitation of Br3− in the presence of EV2+ (EV2+/Br3−; blue open-circle curve vs EV2+/Br3−/MEP+; black solid curve), which correlates with the poor stability and fast self-discharge results obtained for the previous asymmetric KBr/MEPBr cell. This relative interaction

Figure 8. A 1.2 M EVBr2/0.12 M TBABr/2.88 M NaBr cell. (a) Specific energy and cycling stability of EV/Br cells with and without TBABr; the inset shows GCD profiles at the 1000th cycle. (b) Long-term cycling stability of the EV/TBA/Br cell cycled at 2 A/g to 1.3 V, and to 1.35 V; the inset shows GCD profiles for the last three cycles. (c) Open-circuit voltage for 6 h after charging the cell to 1.35 V (orange curve) and 1.3 V (blue curve). (d) Specific energy vs power of the EV/TBA/Br cell (orange curve) and its comparison to Na2SO4 (black curve) and commercial Maxwell (green curve) cells. Specific energy and power of all cells, including the commercial EDLC, were normalized to the dry mass of both positive and negative electrodes. Rate tests of the EV/TBA/Br cell were performed after the initial 1000 GCD cycles at 1 A/g. 9991

DOI: 10.1021/jacs.7b04603 J. Am. Chem. Soc. 2017, 139, 9985−9993

Article

Journal of the American Chemical Society Table 2. Comparison of Selected Previous Reports of Aqueous Redox ECsa electrode

redox electrolyte

specific energy at W/kg (or A/g)

cycling stability

AC

SnSO4/VOSO4

75 W·h/kg (at 0.055 A/g)

85% (4500 cycles)

AC

EVBr2/TBABr/ NaBr

64 W·h/kg (∼25 W·h/L)b (at 1 A/g)

AC

46 W·h/kg (at 1 A/g)

reduced GO functionalized CNT AC

SnF2/VOSO4/ H2SO4 KI/KOH KBr/Na2SO4

90% (5000 cycles to 1.35 V) 96% (7000 cycles to 1.3 V) 80% (6500 cycles)

44 W·h/kg (at 0.83 A/g) 28.3 W·h/kg (at 372 W/kg)

∼80% (5000 cycles) 86.3% (10 000 cycles)

K3Fe(CN)6

28.3 W·h/kg (at 0.05 A/g)

80% (9000 cycles)

AC

EVBr2/H2SO4

23 W·h/kg (at 0.25 A/g)

graphene hydrogel

hydroquinone/ H2SO4 CuSO4/H2SO4

8.9 W·h/kgc (at 252 W/kgc)

no degradation (1000 cycles) 80% (1000 cycles)

6.7 W·h/kgc (at 360 W/kgc)

∼99% (1000 cycles)

separator

self-discharge rate

ref

anion exchange membrane filter paper

1.4 V → 1.16 V (10 h)

19

1.35 V → 1.13 V (6 h)

this work

anion exchange membrane polypropylene cation exchange membrane cation exchange membrane polypropylene

1.4 V → 1.2 V (10 h)

23

N/A N/A

31 28

0.8 V → 0.65 V (10 h)

21

N/A

50

Nafion 117 membrane

0.8 V → 0.3 V (∼1.5 h)

24

porous cellulose acetate

49%c of initial energy (1 h)

a The data are arranged by descending specific energy reported without considering rates. GO = graphene oxide; CNT = carbon nanotube; N/A = not available. bVolumetric energy density was determined to compare our system with selected redox flow batteries that utilize relevant redox couples (e.g., metal-free organic molecules and/or halides) in aqueous systems.47,49,51,52 The comparison is summarized in Table S4. cThe values are not explicitly reported in the paper, and estimated numbers are from the previously reported literature.26

bromine within the pores of the carbon electrodes. This mechanism suppresses unwanted chemical reactivity and crossdiffusion of Br2/Br3− that results in improved cycling stability and self-discharge rates of redox ECs. We used this fundamental understanding of chemical and electrochemical processes at the electrolyte/electrode interface to develop a high-performance dual-redox EC. This dual-redox EC, which integrates TBABr with an ethyl viologen anolyte and a bromide catholyte, produces a record specific energy with stable cycling at high power levels: Addition of tetrabutylammonium improves all aspects of device performance including stability, energy, and power density without adding complexity to the fabrication process. The system can be assembled without a dry room or glovebox, uses aqueous electrolyte with readily available salts, and has a cost-effective paper separator that avoids the use of expensive ion-selective membranes. These attributes address hurdles to practical commercialization, and low costs of production may be possible. Our systematic approach to efficiently utilize bromide catholyte with a non-redox-active cation and design of a high-performance dual-redox EC should be informative and promise to be readily applicable for a wide variety of energy storage applications, especially for the science of hybrid electrochemical energystorage systems (e.g., pairing a high-performance bromide catholyte at the positive electrode with a faradaic battery-type or pseudocapacitive negative electrode).

TBA/Br cell shows good self-discharge characteristics. The EV/ Br control cell has ηR(6 h) = 28% when charged to 1.35 V, due to gradual cross-diffusion of the charged Br3− and EV•+ species from the positive and negative electrodes, respectively. In contrast, the EV/TBA/Br cell, which is designed to suppress the Br−/ Br3− shuttle, shows improved self-discharge with ηR(6 h) = 53% when charged to 1.3 V and ηR(6 h) = 45% when charged to 1.35 V. For these two conditions the open-circuit voltage losses are only 13% and 16%, respectively, comparable to other redox ECs that use costly ion-selective membrane separators (Figure 8c; Table 2). The ability to forego the use of the ion-selective membrane is made possible by the solid-state confinement of the bromine species in the porous carbon electrode. The Ragone plot in Figure 8d shows that the EV/TBA/Br cell provides a power performance higher than 3 kW/kg while still retaining ∼15 W·h/kg energy density. This cell’s energy, power, stability, and self-discharge performance are compared with previously reported values for aqueous redox ECs in Table 2. In order to provide further performance comparison, an aqueous EDLC (1 M Na2SO4 electrolyte) and a commercial nonaqueous EDLC (BCAP0001 P270 T01, Maxwell Technologies) were tested as control/comparison devices. Figure 8d shows that the EV/TBA/Br cell produced specific energy well above the 1 M Na2SO4 cell (black curve) at all power rates. Compared to the commercial nonaqueous cell, this cell delivered as high as ∼250% increased specific energy over a wide range of power rates up to ∼3 kW/kg. Overall, these results further demonstrate that the EV/TBA/Br redox EC system substantially improves energy as well as power densities of aqueous EDLCs by adding dual faradaic reactions at both the positive and negative electrodes, without negatively impacting cycle life.



ASSOCIATED CONTENT

* Supporting Information S

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b04603. Materials, Figures S1−S6, Tables S1−S4, and experimental details including calculations and cell construction (PDF)



CONCLUSION We propose a paradigm shift for the conventional use of a mobile, liquid state to complex Br3− and report a fundamentally different solution to complex/store Br3− for aqueous redox ECs. We demonstrate with microscopic-level evidence that reversible solid complexation of Br−/Br3− redox couple induced by tetrabutylammonium bromide effectively stores reactive and diffusive



AUTHOR INFORMATION

Corresponding Authors

*[email protected] *[email protected] 9992

DOI: 10.1021/jacs.7b04603 J. Am. Chem. Soc. 2017, 139, 9985−9993

Article

Journal of the American Chemical Society ORCID

(23) Lee, J.; Tolosa, A.; Krüner, B.; Jäckel, N.; Fleischmann, S.; Zeiger, M.; Kim, D.; Presser, V. Sustainable Energy Fuels 2017, 1, 299. (24) Chen, L.; Bai, H.; Huang, Z.; Li, L. Energy Environ. Sci. 2014, 7, 1750. (25) Wang, B.; Maciá-Agulló, J. A.; Prendiville, D. G.; Zheng, X.; Liu, D.; Zhang, Y.; Boettcher, S. W.; Ji, X.; Stucky, G. D. J. Electrochem. Soc. 2014, 161, A1090. (26) Chun, S.-E.; Evanko, B.; Wang, X.; Vonlanthen, D.; Ji, X.; Stucky, G. D.; Boettcher, S. W. Nat. Commun. 2015, 6, 7818. (27) Senthilkumar, S. T.; Selvan, R. K.; Lee, Y. S.; Melo, J. S. J. Mater. Chem. A 2013, 1, 1086. (28) Tang, X.; Lui, Y. H.; Chen, B.; Hu, S. J. Power Sources 2017, 352, 118. (29) Li, Q.; Haque, M.; Kuzmenko, V.; Ramani, N.; Lundgren, P.; Smith, A. D.; Enoksson, P. J. Power Sources 2017, 348, 219. (30) Zhao, Y.; Ding, Y.; Song, J.; Peng, L.; Goodenough, J. B.; Yu, G. Energy Environ. Sci. 2014, 7, 1990. (31) Sankar, K. V.; Kalai Selvan, R. Carbon 2015, 90, 260. (32) Biswas, S.; Senju, A.; Mohr, R.; Hodson, T.; Karthikeyan, N.; Knehr, K. W.; Hsieh, A. G.; Yang, X.; Koel, B. E.; Steingart, D. A. Energy Environ. Sci. 2017, 10, 114. (33) Gerhardt, M. R.; Tong, L.; Gómez-Bombarelli, R.; Chen, Q.; Marshak, M. P.; Galvin, C. J.; Aspuru-Guzik, A.; Gordon, R. G.; Aziz, M. J. Adv. Energy Mater. 2017, 7, 1601488. (34) Lancry, E.; Magnes, B.-Z.; Ben-David, I.; Freiberg, M. ECS Trans. 2013, 53, 107. (35) Poon, G.; Parasuraman, A.; Lim, T. M.; Skyllas-Kazacos, M. Electrochim. Acta 2013, 107, 388. (36) Evanko, B.; Yoo, S. J.; Chun, S.-E.; Wang, X.; Ji, X.; Boettcher, S. W.; Stucky, G. D. J. Am. Chem. Soc. 2016, 138, 9373. (37) Hoshino, K.; Ando, M.; Oikawa, Y.; Okuma, M.; Murashiro, K. Sol. Energy Mater. Sol. Cells 2015, 137, 15. (38) Fic, K.; Lota, G.; Frackowiak, E. Electrochim. Acta 2010, 55, 7484. (39) Fic, K.; Lota, G.; Frackowiak, E. Electrochim. Acta 2012, 60, 206. (40) Chen, X.; Rickard, M. A.; Hull, J. W.; Zheng, C.; Leugers, A.; Simoncic, P. Inorg. Chem. 2010, 49, 8684. (41) Zambounis, J. S.; Kamitsos, E. I.; Patsis, A. P.; Papavassiliou, G. C. J. Raman Spectrosc. 1992, 23, 81. (42) Klein, N. D.; Hurley, K. R.; Feng, Z. V.; Haynes, C. L. Anal. Chem. 2015, 87, 4356. (43) Bals, S.; Kabius, B.; Haider, M.; Radmilovic, V.; Kisielowski, C. Solid State Commun. 2004, 130, 675. (44) Genovese, M.; Jiang, J.; Lian, K.; Holm, N. J. Mater. Chem. A 2015, 3, 2903. (45) Gaier, J. R.; Ditmars, N. F.; Dillon, A. R. Carbon 2005, 43, 189. (46) Izumi, I.; Sato, J.; Iwashita, N.; Inagaki, M. Synth. Met. 1995, 75, 75. (47) Janoschka, T.; Martin, N.; Hager, M. D.; Schubert, U. S. Angew. Chem., Int. Ed. 2016, 55, 14427. (48) Hoshino, K.; Oikawa, Y.; Sakabe, I.; Komatsu, T. Electrochim. Acta 2009, 55, 165. (49) Liu, T.; Wei, X.; Nie, Z.; Sprenkle, V.; Wang, W. Adv. Energy Mater. 2016, 6, 1501449. (50) Sathyamoorthi, S.; Kanagaraj, M.; Kathiresan, M.; Suryanarayanan, V.; Velayutham, D. J. Mater. Chem. A 2016, 4, 4562. (51) Hu, B.; DeBruler, C.; Rhodes, Z.; Liu, T. L. J. Am. Chem. Soc. 2017, 139, 1207. (52) Huskinson, B.; Marshak, M. P.; Suh, C.; Er, S.; Gerhardt, M. R.; Galvin, C. J.; Chen, X.; Aspuru-Guzik, A.; Gordon, R. G.; Aziz, M. J. Nature 2014, 505, 195.

Xingfeng Wang: 0000-0003-4992-7993 Xiulei Ji: 0000-0002-4649-9594 Shannon W. Boettcher: 0000-0001-8971-9123 Galen D. Stucky: 0000-0002-0837-5961 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This research was supported by the Advanced Research Projects Agency-Energy (ARPA-E), U.S. Department of Energy (Award No. DE-AR0000344). The MRL Shared Experimental Facilities are supported by the MRSEC Program of the NSF, a member of the NSF-funded Materials Research Facilities Network (www. mrfn.org), under Award No. DMR 1121053. We are grateful to Prof. Martin Moskovits for useful insight and discussion. We thank Dr. Binghui Wu and Tracy T Chuong for assistance with TEM and Raman measurements, respectively, and Alex Schrader for assistance with contact angle measurements. S.W.B. further thanks the Alfred P. Sloan Foundation for support as a Sloan Research Fellow.



REFERENCES

(1) Béguin, F.; Presser, V.; Balducci, A.; Frackowiak, E. Adv. Mater. 2014, 26, 2219. (2) Simon, P.; Gogotsi, Y. Nat. Mater. 2008, 7, 845. (3) Burke, A. J. Power Sources 2000, 91, 37. (4) Blomquist, N.; Wells, T.; Andres, B.; Bäckström, J.; Forsberg, S.; Olin, H. Sci. Rep. 2017, 7, 39836. (5) Wang, Y.; Song, Y.; Xia, Y. Chem. Soc. Rev. 2016, 45, 5925. (6) Fic, K.; Meller, M.; Menzel, J.; Frackowiak, E. Electrochim. Acta 2016, 206, 496. (7) Gao, Q.; Demarconnay, L.; Raymundo-Pinero, E.; Beguin, F. Energy Environ. Sci. 2012, 5, 9611. (8) Arbizzani, C.; Biso, M.; Cericola, D.; Lazzari, M.; Soavi, F.; Mastragostino, M. J. Power Sources 2008, 185, 1575. (9) Augustyn, V.; Simon, P.; Dunn, B. Energy Environ. Sci. 2014, 7, 1597. (10) Sheberla, D.; Bachman, J. C.; Elias, J. S.; Sun, C. J.; Shao-Horn, Y.; Dinca, M. Nat. Mater. 2017, 16, 220. (11) Brezesinski, T.; Wang, J.; Tolbert, S. H.; Dunn, B. Nat. Mater. 2010, 9, 146. (12) Lang, X.; Hirata, A.; Fujita, T.; Chen, M. Nat. Nanotechnol. 2011, 6, 232. (13) Yan, J.; Wang, Q.; Wei, T.; Fan, Z. J. Adv. Energy Mater. 2014, 4, 1300816. (14) Muller, G. A.; Cook, J. B.; Kim, H.-S.; Tolbert, S. H.; Dunn, B. Nano Lett. 2015, 15, 1911. (15) Boota, M.; Chen, C.; Becuwe, M.; Miao, L.; Gogotsi, Y. Energy Environ. Sci. 2016, 9, 2586. (16) Fic, K.; Meller, M.; Frackowiak, E. J. Electrochem. Soc. 2015, 162, A5140. (17) Frackowiak, E.; Meller, M.; Menzel, J.; Gastol, D.; Fic, K. Faraday Discuss. 2014, 172, 179. (18) Roldán, S.; Blanco, C.; Granda, M.; Menéndez, R.; Santamaría, R. Angew. Chem., Int. Ed. 2011, 50, 1699. (19) Lee, J.; Kruner, B.; Tolosa, A.; Sathyamoorthi, S.; Kim, D.; Choudhury, S.; Seo, K.-H.; Presser, V. Energy Environ. Sci. 2016, 9, 3392. (20) Senthilkumar, S. T.; Selvan, R. K.; Melo, J. S. J. Mater. Chem. A 2013, 1, 12386. (21) Lee, J.; Choudhury, S.; Weingarth, D.; Kim, D.; Presser, V. ACS Appl. Mater. Interfaces 2016, 8, 23676. (22) Wang, X.; Chandrabose, R. S.; Chun, S.-E.; Zhang, T.; Evanko, B.; Jian, Z.; Boettcher, S. W.; Stucky, G. D.; Ji, X. ACS Appl. Mater. Interfaces 2015, 7, 19978. 9993

DOI: 10.1021/jacs.7b04603 J. Am. Chem. Soc. 2017, 139, 9985−9993